When Land-Use Models Assume Infinite Patience: The Ethics of Temporal Discounting

You're staring at a model output. It says convert that forest to farmland now, because the net present value is higher. But something gnaws at you. The model assumes a discount rate—say 5%—meaning a dollar's worth of ecosystem services in 50 years is worth pennies today. Is that right? Ethically, it's a minefield.

Land-use models often treat time like a bank account: future costs and benefits get shrunk by a constant factor. But land isn't money. Communities, species, carbon cycles—they don't discount. This piece digs into the ethical knots of temporal discounting in land-use modeling. No neat answers, but better questions.

1. Field context: where discounting shows up in real work

Urban sprawl and infrastructure planning

The road planners I've worked with rarely talk about discounting. They talk about pavement condition indexes, traffic projections, and bond ratings. But every time a city approves a new subdivision on farmland thirty miles out, temporal discounting is the real decision engine. The developer sees a ten-year payoff. The municipality sees three decades of property tax revenue. What gets discounted off the ledger? The eventual cost of repairing that road, extending water lines, and running school buses at 5:30 AM. That sounds fine until the bond comes due and the town realizes it sold a piece of its future for a fraction of today's count. The ethical problem is not ignorance — it's the quiet assumption that a dollar spent fixing infrastructure thirty years from now matters less than the same dollar spent today. Most people reject that logic if you state it bluntly. Yet the models encode it.

Conservation vs. development trade-offs

Conservation groups face a brutal mirror image. If a wetland provides flood protection, carbon storage, and biodiversity habitat across seventy years, a standard net-present-value calculation asks: is that worth more than a one-time timber sale or a housing development? The math usually says no — because the benefits that arrive decades from now get cut so deeply by the discount rate that they barely register. I have sat in meetings where a county commissioner said, 'Be realistic — those future values are imaginary.' That hurts. Because it's partly true: we can measure the timber sale today but only guess at the wetland's value in 2075. The catch is that discounting the future away doesn't make the costs vanish — it just who pays. The developer wins now; the community pays later in flood damage, higher water bills, lost pollinators. Traded across time: cheap today, costly tomorrow.

Discounting assumes the future has a weaker voice. But the future doesn't delegate its vote — it shows up with a bill.

— field note from a land-use planning workshop, 2023

Forestry and agricultural rotation decisions

The odd part is how plantation forestry handles this. A pine rotation runs twenty-five to thirty-five years. That's a single decision cycle — not multiple generations. Pick the wrong discount rate and you either cut too early (losing wood volume and quality) or wait too long (missing a market window). I watched a small family timber operation in the Pacific Northwest run the same spreadsheet three different ways: one at 4%, one at 6%, one at 8%. The cut-year recommendation shifted by eleven years. Eleven years of growth, eleven years of interest payments, eleven years of counting on the same person being alive to sign the deed. The ethical line wobbles here: is it irresponsible to ask a sixty-year-old farmer to plant trees that will mature after retirement? Or is it more irresponsible to burn the option early for cash that could be earned another way? What usually breaks first is the ability to trust a number — the discount rate — that no one can actually observe. It's assumed, debated, then typed into a cell. And then the forest comes down.

2. Foundations readers confuse: discount rate vs. time preference vs. opportunity cost

Pure time preference — not just impatience

Most teams treat discounting as a single knob: turn it up for short-term thinking, down for long-term vision. That oversimplification breaks models. Pure time preference is the human bias to want something now rather than later, even when later offers identical value. It's irrational by strict utility math — yet you see it in every land-use negotiation. A developer pushes for ten-year timber revenue because waiting thirty years feels like waiting for a geological epoch. The catch: pure time preference has zero connection to uncertainty, inflation, or investment opportunity. It's naked impatience. One planner I worked with called it 'the toddler tax on future generations.' That hurts because it's not wrong.

Models encode this as a flat percentage added to the discount rate. Standard practice uses 0.5–2 % above the risk-free rate. What usually breaks first is the assumption that this preference stays constant across time horizons. It doesn't. People tolerate waiting for a house they will live in; they demand immediate return on a conservation offset they will never see. Pure time preference behaves like a leaky bucket — the farther out the benefit, the more the bias grows. Ignoring that skews any simulation past twenty years into junk.

Marginal utility and growth — the wealth trap

A dollar today buys more than a dollar tomorrow — but only if you're poor today. That's the logic of marginal utility: as societies grow richer, each additional unit of consumption yields less well-being. The standard assumption says future generations will be wealthier, so their gains matter less in present value. This is where ethical trouble hides. If your growth projection overshoots — say you assume 2.5 % annual GDP growth when a region stagnates at 0.9 % — then your model systematically devalues resources that future communities will actually need to survive. I have seen this destroy a wetland restoration proposal: the discount rate reflected global growth averages, but the local watershed was losing income. The model said 'protecting it costs too much.' The river said otherwise.

Marginal utility arguments also assume consumption elasticity stays stable. Wrong order. A food-insecure population values an extra bushel of grain far more than a wealthy one does. When your land-use model applies a single growth curve across diverse income groups, you're not simplifying — you're erasing distributional ethics. The trade-off: higher growth assumptions make long-term projects look unattractive; lower assumptions make them look urgent. Neither is neutral.

Uncertainty and risk premium — the easy scapegoat

Risk premium is the number you add to a model when you don't want to defend the real source of your doubt.

— overheard at a planning review, state forestry office

Uncertainty feels like a safe place to hide. Future carbon prices? Unknown. Regulatory shifts? Unpredictable. So teams tack on 2–3 % risk premium and call it conservative. That's lazy. Risk premium properly reflects the variance of possible outcomes, not the absence of data. If you have no clue about future land-use policy, adding a spread to the discount rate doesn't fix the model — it hides the ignorance behind a decimal. What actually helps: running Monte Carlo on key uncertain variables and letting the rate float with scenario ranges. One team I advised dropped a fixed 4 % premium and switched to a band of 1.5–6 % depending on political stability scores. Their forest conversion projections stopped jumping by 40 % every election cycle.

Honestly — most urban posts skip this.

The odd part is — risk premium routinely conflates two separate things: the chance that the future payoff never arrives, and the chance that the payoff arrives but is worth less than expected. These demand different treatments. Default risk leans on insurance logic; value uncertainty leans on option pricing. Most land-use models ignore the distinction. That leads to double-counting or, worse, ethical sleight-of-hand: jacking up the rate because 'the future is uncertain,' then using the inflated rate to justify short-term extraction. A neat trick — if you don't mind pretending uncertainty always argues for consumption now.

3. Patterns that usually work: when discounting aligns with ethics

Short-term decisions with reversible impacts

The easy cases aren't sexy, but they keep projects honest. When you're choosing between two paving materials for a parking lot that will be ripped up in five years anyway, discounting at a standard rate works fine. The key: reversibility. A temporary gravel lot, a seasonal pop-up market, a one-season crop rotation—these decisions carry low stakes. I have seen teams agonize for weeks over discount rates for a three-year lease, then miss the lease deadline entirely. That hurts more than getting the rate slightly wrong. The ethical move here is to match your discounting horizon to the physical lifespan of whatever you're building. If you can undo the decision in your lifetime, pick a rate and move on.

High discount rates for private investment

Private capital behaves like a sprinter, not a marathoner—and that's fine for certain land uses. A developer building a 15-story apartment tower on a 50-year lease can ethically apply a high discount rate because the investor's timeline is finite. The odd part is—the same developer should use a different rate for the ground-floor public plaza they promised the city. Different beneficiaries, different ethics. The pitfall: many teams apply one rate to the entire project budget, blurring private gain and public obligation. Separate the streams. Use 8–12% on the revenue-generating floors. Use 1–3% on the sidewalk widening. Nobody complains when the returns are transparent.

'Discounting is morally neutral—it becomes unethical only when we pretend the future has no voice in the room.'

— paraphrased from a zoning board member in a coastal city, 2023

Declining discount rates for long-term public goods

This is where the math meets a conscience. Standard exponential discounting treats year 50 the same way it treats year 5—which is absurd for things like wetlands, aquifer recharge zones, or urban forests. A declining discount rate says: use 5% for years 1–10, 3% for years 11–30, and 1% for years 31–100. The logic? Short-term opportunity cost is real; long-term social obligation is also real. We fixed this on a coastal resilience plan by showing the city council that a straight 6% rate made the sea wall look pointless—when everyone knew the sea wall was necessary. The declining schedule flipped the outcome. That said, declining rates can invite gaming. Teams sometimes inflate the early rate to justify bad short-term choices. Wrong order. You build the declining schedule first, then run the numbers through it—not the reverse. Most teams skip this step, then wonder why the 80-year tree canopy program gets axed. The seam blows out because the discount rate lied about the future.

What usually breaks first is the assumption that the same rate works for asphalt and aquifers. It doesn't. Pick the rate after you decide whose time horizon counts—the bondholder's or the watershed's.

4. Anti-patterns and why teams revert

Using a single constant rate for all impacts

The most seductive mistake is choosing one discount rate—say, 4%—and applying it to every consequence in the model. That sounds fine until you realize you've treated a short-term carbon credit and a century-scale aquifer depletion as identical math problems. They aren't. A flat rate compresses distant damages so aggressively that a catastrophic flood in 2080 registers as a minor cost, barely worth a footnote in the net-present-value column. Most teams revert because the single-rate method is fast, it's what the spreadsheet template expects, and arguing about differentiated rates opens a political can of worms no project manager wants to touch. But speed here buys a false precision that misleads decision-makers about what actually matters.

The odd part is—teams often know this is broken. I have watched modelers shrug and say "we'll adjust later," only to never return. The fix isn't elegant, but it works: split your impacts into three baskets—reversible near-term, persistent medium-term, and irreversible long-term—and assign separate declining-rate schedules to each. That hurts the neatness of the output table. Good. Ugly honesty beats polished deception.

Ignoring intergenerational equity

Discounting inherently favors the present. Apply a 5% rate over sixty years and a single dollar of harm to your grandchildren's generation gets reduced to five cents today. That's a value judgment dressed up as arithmetic. Many modelers sidestep it entirely: they use the firm's cost of capital as their discount proxy and call the equity question "somebody else's job." But somebody else never shows up. The result is a land-use plan that quietly justifies extracting short-term profit while dumping long-term restoration costs onto people who aren't yet born to protest.

‘The future is discounted at a rate that makes it vanish. The present is all that survives the math.’

— overheard in a planning debrief, after a model showed zero cost for a 2095 soil collapse

What usually breaks first is trust. Communities reviewing the model spot the asymmetry immediately—they don't need a PhD to see that their children's water supply got written down to pennies. Teams revert to bad defaults because intergenerational equity feels like a philosophical distraction from the "real" work of hitting quarterly targets. Wrong order. The real work is admitting that every discount rate encodes an ethical stance. Name it explicitly, or let the default rate name it for you—and the default is almost always selfish.

Not every urban checklist earns its ink.

Discounting irreversible losses

Losses that can't be undone—species extinction, ancient soil structure, cultural landscapes—should not be discounted at all. You can't "replace" a lost language grove or regenerate a thousand-year-old peat dome within any plausible modelling horizon. Yet standard practice applies the same compounding knife to these endpoints, shaving their present value until they fall below the cost of mitigation. That's not analysis; it's a loophole that excuses inaction. Teams revert to this anti-pattern because the alternative—flagging an impact as infinite or priceless—breaks the tidy cost-benefit framework their clients demand.

The catch is that pretending priceless things have a price doesn't make them affordable; it just hides the debt. I have seen a team patch this with a simple boundary rule: any impact flagged as irreversible triggers a separate qualitative overlay, removed from the discounted spreadsheet entirely. No discount formula touches it. That overlay then forces a human judgement call, not a number-crunching subroutine. It slows the process, introduces friction, and occasionally kills profitable projects. That's precisely why most teams resist it—and precisely why you should do it anyway. The next chapter deals with what happens when the model drifts and those hard-won boundaries begin to erode.

5. Maintenance, drift, or long-term costs

Rate updating over time

The discount rate you set in year one is almost certainly wrong by year ten. That sounds like a planning failure — it's actually a feature. Land-use models that treat discounting as a permanent calibration (pick 4%, done) ignore how social preferences shift across decades. I have watched teams cling to a single rate through two economic crises, insisting the number was "peer-reviewed" and therefore sacred. The problem: the peer review was done under different climate expectations, different political stability, different migration pressures. The rate becomes a fossil — still shaped like a discount, but no longer alive. You fix this by scheduling rate reviews at fixed intervals, but even that creates friction. Every review invites political actors to argue for a lower rate (more conservation, less development) or a higher one (stimulus, revenue, now). The moral math gets messy fast.

The catch is that updating feels like admitting failure. Most governance documents skip this step. They define a procedure for picking the initial rate, then stay silent on revision. That silence breeds drift — the model keeps running, but its ethical assumptions quietly fossilize.

Political pressure to adjust rates

Discount rates don't live in a neutral spreadsheet. They're wielded by agencies, councils, and planning boards that face election cycles and budget fights. I have seen a perfectly reasonable 3.5% rate suddenly spiral to 7% because a new administration wanted to fast-track a transit corridor. The justification sounded technical — "We adjusted for liquidity preference" — but the actual driver was a four-year time horizon, not a seventy-year land ethic.

What usually breaks first is the boundary between technical revision and political manipulation. Good governance draws this line sharp: rate changes must follow pre-registered criteria, not quarterly priorities. Bad governance — the kind I see most often — embeds the discount rate in a political approval process, meaning the rate becomes a bargaining chip. The ethical cost is deferred to the population who will live with the land-use outcome, but they don't sit at the bargaining table. Wrong order.

'We treat the discount rate as a thermostat, not a steering wheel. A thermostat you set once and trust. A steering wheel you grab every time the road bends.'

— overheard at a land-use governance workshop, 2023

Data obsolescence and recalibration

Even if the rate stays fixed, the data it operates on decays. Land productivity, water availability, population density — all shift. A 2020 suitability map doesn't track 2030's drought patterns. The model spits out elegant curves, but the input numbers are ghosts. Recalibration is expensive. It requires new field surveys, updated remote sensing, stakeholder re-engagement. Most planning departments don't budget for this. They budget for the model build, not its maintenance over a human lifetime.

That hurts. When the gap between stale data and real conditions widens, the discounting logic becomes unethical — not because the math is wrong, but because the world has moved while the model stayed still. One fix: tag every input with an expiration date. Nine years out, the model flags the variable as degraded. That forces a conversation instead of a silent continuation. Not yet standard practice, but it should be.

6. When not to use this approach

Irreversible biodiversity loss

Discounting works fine when you can reverse a decision next year. You clear a field, plant soy for five seasons, then let it fallow—fine. Wrong order for an ancient woodland. When a land-use model applies a standard 5% discount rate to a habitat that took four centuries to assemble, it says, effectively: That old-growth structure is worth pennies today. The catch is—once that soil profile, that mycorrhizal web, that canopy structure is gone, you don't get it back. I have seen teams run carbon-offset models that happily swap a primary forest for a monoculture plantation because the spreadsheet showed net present value in the black. The model wasn't wrong. It was inappropriate. The boundary here is simple: if the ecological asset can't be rebuilt within a human generation, discounting is not just a technical choice—it's an ethical decision dressed in math.

Long time horizons beyond 100 years

Discount rates collapse the future. At a 3% discount rate, one dollar of damage two hundred years from now is worth less than one cent today. That sounds like a harmless abstraction. It's not. We're deciding, right now, whether to store nuclear waste, whether to build coastal defenses that lock in shoreline loss for centuries, whether to drain peatlands that would take millennia to regenerate. The model treats those distant costs as noise. Most teams skip this: they calibrate their discount rate using today's capital markets, then apply it to a 500-year simulation. That's a category error. The opportunity cost of capital tells you what you could have earned investing elsewhere—it doesn't tell you how to weigh the survival of a coastal city in 2150. For horizons beyond 100 years, I prefer zero discounting or a declining-rate schedule. Not because it's mathematically elegant. Because it forces the conversation to stay moral rather than mechanical.

Discounting at 3% over 300 years makes a million-dollar catastrophe look like a $4 parking ticket today.

— paraphrase from a restless modeler at a land-use conference, 2023

Reality check: name the planning owner or stop.

The odd part is—teams that resist long-horizon discounting often feel unscientific. They worry they're abandoning rigor. I think the opposite: applying a constant discount rate across centuries is the lazy move. It lets you avoid the hard question of what we owe people we will never meet. A simple test: if the simulation's outcome changes drastically by shifting discount rates from 2% to 4%, you're not measuring land-use dynamics—you're measuring the discount rate. Stop trying the same method harder. Switch to scenario analysis without discounting, or use a tiny, fixed absolute weight for each future decade.

Decisions involving catastrophic risk

Here the model breaks in a different way. Standard discounting assumes that you can average outcomes—good years trade off against bad years. That assumption fails catastrophically when one outcome involves the permanent collapse of a water system or the extinction of a keystone species. Discounting smooths these events into a probabilistic footnote. A 1-in-100 chance of aquifer sterilization gets multiplied by the discount factor and buried in the expected value. But when it happens, you don't experience the *average* of all scenarios—you live in the one where the wells ran dry. I fixed this once by splitting the decision tree: run a discounted cash flow for the routine operations, then run a separate no-discount analysis on the catastrophic branch alone. The board hated the extra work. They approved the project anyway, because the black-swan scenario finally looked terrifying without three decades of discounting shrinking it. That's the boundary: if the worst-case outcome destroys the system that the model assumes is stable, discounting is not just unethical—it's unreal. Use hazard-avoidance logic instead. Or, simpler: ask whether you could explain the model's recommendation to somebody living through the disaster. If the answer makes you wince, use a different tool.

7. Open questions / FAQ

Should we use zero discount rate for health impacts?

The short answer is not in isolation — but the debate is real. Some planners argue that human health has no time price: a life saved in year 2100 is identical to a life saved today. That sounds noble until you realize a zero rate can steer investment away from immediate interventions. I have seen project teams freeze when they model a 0% discount for health while everything else — construction costs, land values, inflation — runs at 4–7%. The seam blows out.

The trade-off is brutal. Give health a zero rate and you implicitly assume future technology and baseline wellbeing are irrelevant. A child with malaria today versus a cancer therapy in 2080 — both get equal weight. Most practitioners I know dodge this by separating budget silos: health impacts use a 1.5–2% social rate while infrastructure runs at 3.5%. Does that make ethical sense? No. But it prevents paralysis.

‘Zero discount on health alone creates perverse incentives: build nothing now, save everyone in the far future.’

— city sustainability officer, unpublished workshop notes

How to incorporate uncertainty about future generations' preferences?

We can't. Honest answer. Nobody knows if people in 2120 will value old-growth forests over dense housing, or prefer air quality to energy independence. The temptation is to assume they will think like us — a safe move that papers over the unknown. Wrong order.

What usually breaks first is the discount schedule itself. Teams adopt a declining rate: high for the first 30 years, then tapering. The logic: early costs are concrete, far-future benefits are speculative. That fix passes ethical muster better than a flat rate, but it introduces a new snag — you need to justify the taper slope. I once watched a policy group spend six meetings arguing whether year 50 should be 2.2% or 1.8%, while the actual land-use decisions sat frozen for eighteen months.

A simpler path exists. Instead of guessing preferences, model multiple discount trajectories and flag where the ethical tension flips. Example: if a reforestation plan shows net loss under 3% but net gain under 1.5%, that tells you the decision is not about arithmetic — it's about whose time horizon counts. Surface that tension rather than bury it under a single rate.

What role for ethical deliberation in model design?

The odd part is: most teams relegate ethics to a closing slide in the executive summary. That's backward. The discount rate is the ethics — it encodes whose wellbeing matters, and when. Ethical deliberation should happen before you type a single formula, not after the sensitivity charts look ugly.

Practical move: run a ‘premortem’ session before the model is built. Gather the stakeholders — planner, economist, community rep, ecologist — and ask: What discount choice would we be ashamed to explain to the public? The answers are uneven but clarifying. Some will say ‘anything above 5% degrades future lives.’ Others will say ‘below 2% starves current housing.’ Neither is right. Both force the modeler to surface assumptions rather than hide them in a parameter.

One caution: ethical deliberation can stall decision-making if it becomes open-ended. Set a time-box — three hours, not three months — and commit to documenting which value judgments the chosen rate represents. That documentation becomes the model’s ethical appendix. Next person who picks up the file can see the debate, not just the decimal.